Face Generation¶

In this project, you'll use generative adversarial networks to generate new images of faces.

Get the Data¶

You'll be using two datasets in this project:

MNIST
CelebA

Since the celebA dataset is complex and you're doing GANs in a project for the first time, we want you to test your neural network on MNIST before CelebA. Running the GANs on MNIST will allow you to see how well your model trains sooner.

If you're using FloydHub, set data_dir to "/input" and use the FloydHub data ID "R5KrjnANiKVhLWAkpXhNBe".

data_dir = './data'

# FloydHub - Use with data ID "R5KrjnANiKVhLWAkpXhNBe"
#data_dir = '/input'


"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import helper

helper.download_extract('mnist', data_dir)
helper.download_extract('celeba', data_dir)

Found mnist Data
Found celeba Data

Explore the Data¶

MNIST¶

As you're aware, the MNIST dataset contains images of handwritten digits. You can view the first number of examples by changing show_n_images.

## I had to move this section here so I don't have to pull the data again
data_dir = './data'
import helper
import numpy as np
show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
%matplotlib inline
import os
from glob import glob
from matplotlib import pyplot

mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'mnist/*.jpg'))[:show_n_images], 28, 28, 'L')
pyplot.imshow(helper.images_square_grid(mnist_images, 'L'), cmap='gray')

<matplotlib.image.AxesImage at 0x7f4b97652278>

CelebA¶

The CelebFaces Attributes Dataset (CelebA) dataset contains over 200,000 celebrity images with annotations. Since you're going to be generating faces, you won't need the annotations. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'img_align_celeba/*.jpg'))[:show_n_images], 28, 28, 'RGB')
pyplot.imshow(helper.images_square_grid(mnist_images, 'RGB'))

<matplotlib.image.AxesImage at 0x7f4b97583828>

Preprocess the Data¶

Since the project's main focus is on building the GANs, we'll preprocess the data for you. The values of the MNIST and CelebA dataset will be in the range of -0.5 to 0.5 of 28x28 dimensional images. The CelebA images will be cropped to remove parts of the image that don't include a face, then resized down to 28x28.

The MNIST images are black and white images with a single color channel while the CelebA images have 3 color channels (RGB color channel).

Build the Neural Network¶

You'll build the components necessary to build a GANs by implementing the following functions below:

model_inputs
discriminator
generator
model_loss
model_opt
train

Check the Version of TensorFlow and Access to GPU¶

This will check to make sure you have the correct version of TensorFlow and access to a GPU

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
from distutils.version import LooseVersion
import warnings
import tensorflow as tf

# Check TensorFlow Version
assert LooseVersion(tf.__version__) >= LooseVersion('1.0'), 'Please use TensorFlow version 1.0 or newer.  You are using {}'.format(tf.__version__)
print('TensorFlow Version: {}'.format(tf.__version__))

# Check for a GPU
if not tf.test.gpu_device_name():
    warnings.warn('No GPU found. Please use a GPU to train your neural network.')
else:
    print('Default GPU Device: {}'.format(tf.test.gpu_device_name()))

TensorFlow Version: 1.0.0
Default GPU Device: /gpu:0

Input¶

Implement the model_inputs function to create TF Placeholders for the Neural Network. It should create the following placeholders:

Real input images placeholder with rank 4 using image_width, image_height, and image_channels.
Z input placeholder with rank 2 using z_dim.
Learning rate placeholder with rank 0.

Return the placeholders in the following the tuple (tensor of real input images, tensor of z data)

import problem_unittests as tests

def model_inputs(image_width, image_height, image_channels, z_dim):
    """
    Create the model inputs
    :param image_width: The input image width
    :param image_height: The input image height
    :param image_channels: The number of image channels
    :param z_dim: The dimension of Z
    :return: Tuple of (tensor of real input images, tensor of z data, learning rate)
    """
    # TODO: Implement Function
    real_input = tf.placeholder(dtype=tf.float32, shape=(None, image_width, image_height, image_channels), name='real_input')
    z_data = tf.placeholder(tf.float32, shape=(None, z_dim), name='z_data')
    lr = tf.placeholder(tf.float32, name='learning_rate')
    
    return real_input, z_data, lr


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_inputs(model_inputs)

Tests Passed

Discriminator¶

Implement discriminator to create a discriminator neural network that discriminates on images. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "discriminator" to allow the variables to be reused. The function should return a tuple of (tensor output of the discriminator, tensor logits of the discriminator).

def discriminator(images, reuse=False):
    """
    Create the discriminator network
    :param images: Tensor of input image(s)
    :param reuse: Boolean if the weights should be reused
    :return: Tuple of (tensor output of the discriminator, tensor logits of the discriminator)
    """
    # TODO: Implement Function
    alpha = 0.01
    
    def layer(val, filt, shape, strides, normalize=True):
        layer = tf.layers.conv2d(val, filt, shape, strides, padding='same')
        if (normalize == True):
            layer = tf.layers.batch_normalization(layer, training=True)
        
        layer = tf.maximum(alpha * layer, layer)
        
        return layer
    
    with tf.variable_scope('discriminator', reuse=reuse):
        
        hl_1 = layer(images, 128, 5, 2, normalize=False)
        hl_2 = layer(hl_1, 256, 5, 2)
        hl_3 = layer(hl_2, 512, 5, 2)
        
        
        flat = tf.reshape(hl_3, (-1, 4*4*512))
        logits = tf.layers.dense(flat, 1)
        
        # passing the fully connected layer through the sigmoid function to make sure
        # it returns one for real image and zero for fake images
        output = tf.sigmoid(logits)

        return output, logits


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_discriminator(discriminator, tf)

Tests Passed

Generator¶

Implement generator to generate an image using z. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "generator" to allow the variables to be reused. The function should return the generated 28 x 28 x out_channel_dim images.

def generator(z, out_channel_dim, is_train=True):
    """
    Create the generator network
    :param z: Input z
    :param out_channel_dim: The number of channels in the output image
    :param is_train: Boolean if generator is being used for training
    :return: The tensor output of the generator
    """
    # TODO: Implement Function
    alpha = 0.01
    
    def layer(val, filt, shape, strides):
        layer = tf.layers.conv2d_transpose(val, filt, shape, strides, padding='SAME')
        layer = tf.layers.batch_normalization(layer, training=is_train)
        layer = tf.maximum(alpha * layer, layer)
        
        return layer
    
    with tf.variable_scope("generator", reuse=not is_train):
        
        h1 = tf.layers.dense(z, 7*7*512)
        h1 = tf.reshape(h1, (-1, 7, 7, 512))
        h1 = tf.layers.batch_normalization(h1, training=is_train)
        h1 = tf.maximum(alpha * h1, h1)
        
        h2 = layer(h1, 256, 5, 1)
        h3 = layer(h2, 128, 5, 2)
    
        logits = tf.layers.conv2d_transpose(h3, out_channel_dim, 5, 2, padding='SAME')
        output = tf.tanh(logits)
        
        
        return output


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_generator(generator, tf)

Tests Passed

Loss¶

Implement model_loss to build the GANs for training and calculate the loss. The function should return a tuple of (discriminator loss, generator loss). Use the following functions you implemented:

discriminator(images, reuse=False)
generator(z, out_channel_dim, is_train=True)

Optimization¶

Implement model_opt to create the optimization operations for the GANs. Use tf.trainable_variables to get all the trainable variables. Filter the variables with names that are in the discriminator and generator scope names. The function should return a tuple of (discriminator training operation, generator training operation).

def model_loss(input_real, input_z, out_channel_dim):
    """
    Get the loss for the discriminator and generator
    :param input_real: Images from the real dataset
    :param input_z: Z input
    :param out_channel_dim: The number of channels in the output image
    :return: A tuple of (discriminator loss, generator loss)
    """
    # TODO: Implement Function
    model = generator(input_z, out_channel_dim)
    
    discriminator_real_model, discriminator_real_logits = discriminator(input_real)
    discriminator_dum_model, discriminator_dum_logits = discriminator(model, True)
    
    # Getting one for real images
    discriminator_r_param = tf.nn.sigmoid_cross_entropy_with_logits(logits=discriminator_real_logits, labels=tf.ones_like(discriminator_real_logits)*np.random.uniform(0.7, 1.2))
    discriminator_real_loss = tf.reduce_mean(discriminator_r_param)
    
    # Getting zero for fake images
    discriminator_d_param = tf.nn.sigmoid_cross_entropy_with_logits(logits=discriminator_dum_logits, labels=tf.zeros_like(discriminator_dum_logits)*np.random.uniform(0.0, 0.3))
    discriminator_dum_loss = tf.reduce_mean(discriminator_d_param)
    
    # The total lost for the discriminator is the sum of lost for the real and the fake images
    discriminator_loss = discriminator_real_loss + discriminator_dum_loss
    
    generator_loss = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=discriminator_dum_logits, labels=tf.ones_like(discriminator_dum_model)))
    
    return discriminator_loss, generator_loss


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_loss(model_loss)

Tests Passed

def model_opt(d_loss, g_loss, learning_rate, beta1):
    """
    Get optimization operations
    :param d_loss: Discriminator loss Tensor
    :param g_loss: Generator loss Tensor
    :param learning_rate: Learning Rate Placeholder
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :return: A tuple of (discriminator training operation, generator training operation)
    """
    # TODO: Implement Function
    
    #get the trainable variables from my Graph
    t_vars = tf.trainable_variables()
    
    # Getting discriminator trainable variables from t_vars
    discriminator_vars = [var for var in t_vars if var.name.startswith('discriminator')]
    
    # Getting generator trainable variables from t_vars
    generator_vars = [var for var in t_vars if var.name.startswith('generator')]
    
    
    with tf.control_dependencies(tf.get_collection(tf.GraphKeys.UPDATE_OPS)):
        d_opt = tf.train.AdamOptimizer(learning_rate, beta1=beta1).minimize(d_loss, var_list=discriminator_vars)
        g_opt = tf.train.AdamOptimizer(learning_rate, beta1=beta1).minimize(g_loss, var_list=generator_vars)
            
        return d_opt, g_opt



"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_opt(model_opt, tf)

Tests Passed

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import numpy as np


def show_generator_output(sess, n_images, input_z, out_channel_dim, image_mode):
    """
    Show example output for the generator
    :param sess: TensorFlow session
    :param n_images: Number of Images to display
    :param input_z: Input Z Tensor
    :param out_channel_dim: The number of channels in the output image
    :param image_mode: The mode to use for images ("RGB" or "L")
    """
    cmap = None if image_mode == 'RGB' else 'gray'
    z_dim = input_z.get_shape().as_list()[-1]
    example_z = np.random.uniform(-1, 1, size=[n_images, z_dim])

    samples = sess.run(
        generator(input_z, out_channel_dim, False),
        feed_dict={input_z: example_z})

    images_grid = helper.images_square_grid(samples, image_mode)
    pyplot.imshow(images_grid, cmap=cmap)
    pyplot.show()

Neural Network Training¶

Show Output¶

Use this function to show the current output of the generator during training. It will help you determine how well the GANs is training.

Train¶

Implement train to build and train the GANs. Use the following functions you implemented:

model_inputs(image_width, image_height, image_channels, z_dim)
model_loss(input_real, input_z, out_channel_dim)
model_opt(d_loss, g_loss, learning_rate, beta1)

Use the show_generator_output to show generator output while you train. Running show_generator_output for every batch will drastically increase training time and increase the size of the notebook. It's recommended to print the generator output every 100 batches.

import matplotlib.pyplot as plt
def train(epoch_count, batch_size, z_dim, learning_rate, beta1, get_batches, data_shape, data_image_mode):
    """
    Train the GAN
    :param epoch_count: Number of epochs
    :param batch_size: Batch Size
    :param z_dim: Z dimension
    :param learning_rate: Learning Rate
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :param get_batches: Function to get batches
    :param data_shape: Shape of the data
    :param data_image_mode: The image mode to use for images ("RGB" or "L")
    """
    # TODO: Build Model
    input_real, input_z, lr = model_inputs(data_shape[1], data_shape[2], data_shape[3], z_dim)
    d_loss, g_loss = model_loss(input_real, input_z, data_shape[3])
    d_opt, g_opt = model_opt(d_loss, g_loss, lr, beta1)
    
    steps = 0
    loss_list = []
    with tf.Session() as sess:
        sess.run(tf.global_variables_initializer())
        for epoch_i in range(epoch_count):
            for batch_images in get_batches(batch_size):
                # TODO: Train Model
                steps += 1
                batch_images *= 2
                batch_z = np.random.uniform(-1, 1, size=(batch_size, z_dim))
                
                _ = sess.run(d_opt, feed_dict={input_real: batch_images, input_z: batch_z, lr: learning_rate})
                _ = sess.run(g_opt, feed_dict={input_real: batch_images, input_z: batch_z, lr: learning_rate})
                
                loss_list.append([d_loss.eval({input_z: batch_z, input_real: batch_images}), g_loss.eval({input_z: batch_z})])
                if steps % print_every == 0:
                    train_loss_d = d_loss.eval({input_z: batch_z, input_real: batch_images})
                    train_loss_g = g_loss.eval({input_z: batch_z})

                    print("Epoch {}/{}...".format(epoch_i + 1, epochs),
                          "Discriminator Loss: {:.4f}...".format(train_loss_d),
                          "Generator Loss: {:.4f}".format(train_loss_g))
                #print(loss_list)
                if steps % show_every == 0:
                    # reusing the show_n_images variable declared and initialized earlier to create the grid
                    show_generator_output(sess, show_n_images, input_z, data_shape[3], data_image_mode)
                    
        plt.plot(loss_list, label='Discriminator')
        plt.plot(loss_list, label='generator')
        plt.ylabel('Loss')
        plt.xlabel('Steps')
        plt.title("Loss Graph")
        plt.show()

MNIST¶

Test your GANs architecture on MNIST. After 2 epochs, the GANs should be able to generate images that look like handwritten digits. Make sure the loss of the generator is lower than the loss of the discriminator or close to 0.

batch_size = 16
z_dim = 16
learning_rate = 0.0002
beta1 = 0.3
print_every = 10
show_every = 100

"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 2

mnist_dataset = helper.Dataset('mnist', glob(os.path.join(data_dir, 'mnist/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, mnist_dataset.get_batches,
          mnist_dataset.shape, mnist_dataset.image_mode)

Epoch 1/2... Discriminator Loss: 1.3047... Generator Loss: 3.1063
Epoch 1/2... Discriminator Loss: 2.3989... Generator Loss: 0.3027
Epoch 1/2... Discriminator Loss: 1.8398... Generator Loss: 4.7896
Epoch 1/2... Discriminator Loss: 2.6739... Generator Loss: 0.1910
Epoch 1/2... Discriminator Loss: 2.2978... Generator Loss: 0.2267
Epoch 1/2... Discriminator Loss: 1.2907... Generator Loss: 0.7616
Epoch 1/2... Discriminator Loss: 1.6401... Generator Loss: 3.4841
Epoch 1/2... Discriminator Loss: 1.0322... Generator Loss: 1.8321
Epoch 1/2... Discriminator Loss: 1.0172... Generator Loss: 1.5472
Epoch 1/2... Discriminator Loss: 1.8311... Generator Loss: 0.4463

Epoch 1/2... Discriminator Loss: 1.4658... Generator Loss: 0.6004
Epoch 1/2... Discriminator Loss: 1.5011... Generator Loss: 0.5885
Epoch 1/2... Discriminator Loss: 1.7142... Generator Loss: 0.5014
Epoch 1/2... Discriminator Loss: 0.9994... Generator Loss: 1.5760
Epoch 1/2... Discriminator Loss: 0.8225... Generator Loss: 2.5631
Epoch 1/2... Discriminator Loss: 0.8632... Generator Loss: 1.5139
Epoch 1/2... Discriminator Loss: 1.1301... Generator Loss: 2.7559
Epoch 1/2... Discriminator Loss: 1.2959... Generator Loss: 0.7076
Epoch 1/2... Discriminator Loss: 1.1412... Generator Loss: 2.4329
Epoch 1/2... Discriminator Loss: 0.8717... Generator Loss: 2.1754

Epoch 1/2... Discriminator Loss: 0.9860... Generator Loss: 1.9360
Epoch 1/2... Discriminator Loss: 1.4978... Generator Loss: 0.5657
Epoch 1/2... Discriminator Loss: 0.9358... Generator Loss: 1.2217
Epoch 1/2... Discriminator Loss: 1.0735... Generator Loss: 0.9763
Epoch 1/2... Discriminator Loss: 1.7580... Generator Loss: 0.4855
Epoch 1/2... Discriminator Loss: 0.9278... Generator Loss: 1.5392
Epoch 1/2... Discriminator Loss: 1.0869... Generator Loss: 0.9998
Epoch 1/2... Discriminator Loss: 1.1417... Generator Loss: 0.9883
Epoch 1/2... Discriminator Loss: 0.8908... Generator Loss: 1.2971
Epoch 1/2... Discriminator Loss: 0.9748... Generator Loss: 2.8153

Epoch 1/2... Discriminator Loss: 1.5856... Generator Loss: 3.6234
Epoch 1/2... Discriminator Loss: 0.9514... Generator Loss: 1.7962
Epoch 1/2... Discriminator Loss: 0.9675... Generator Loss: 2.1978
Epoch 1/2... Discriminator Loss: 0.9384... Generator Loss: 1.4164
Epoch 1/2... Discriminator Loss: 1.4374... Generator Loss: 0.5866
Epoch 1/2... Discriminator Loss: 1.4389... Generator Loss: 3.1402
Epoch 1/2... Discriminator Loss: 1.1998... Generator Loss: 0.8038
Epoch 1/2... Discriminator Loss: 1.0910... Generator Loss: 2.1144
Epoch 1/2... Discriminator Loss: 1.4904... Generator Loss: 3.5163
Epoch 1/2... Discriminator Loss: 1.1464... Generator Loss: 2.1301

Epoch 1/2... Discriminator Loss: 1.3220... Generator Loss: 2.9478
Epoch 1/2... Discriminator Loss: 1.2074... Generator Loss: 0.8046
Epoch 1/2... Discriminator Loss: 1.2509... Generator Loss: 0.9012
Epoch 1/2... Discriminator Loss: 1.1741... Generator Loss: 0.8170
Epoch 1/2... Discriminator Loss: 0.9558... Generator Loss: 1.4574
Epoch 1/2... Discriminator Loss: 1.3786... Generator Loss: 0.7437
Epoch 1/2... Discriminator Loss: 0.9148... Generator Loss: 1.4110
Epoch 1/2... Discriminator Loss: 0.9701... Generator Loss: 1.4583
Epoch 1/2... Discriminator Loss: 1.0647... Generator Loss: 1.3308
Epoch 1/2... Discriminator Loss: 1.0862... Generator Loss: 0.9901

Epoch 1/2... Discriminator Loss: 0.9691... Generator Loss: 1.2265
Epoch 1/2... Discriminator Loss: 1.1871... Generator Loss: 0.8002
Epoch 1/2... Discriminator Loss: 0.8828... Generator Loss: 1.7065
Epoch 1/2... Discriminator Loss: 0.8674... Generator Loss: 1.9281
Epoch 1/2... Discriminator Loss: 0.9320... Generator Loss: 1.4038
Epoch 1/2... Discriminator Loss: 1.1470... Generator Loss: 0.8265
Epoch 1/2... Discriminator Loss: 1.4409... Generator Loss: 0.7195
Epoch 1/2... Discriminator Loss: 1.1014... Generator Loss: 0.9145
Epoch 1/2... Discriminator Loss: 1.1140... Generator Loss: 0.8899
Epoch 1/2... Discriminator Loss: 1.0978... Generator Loss: 2.6823

Epoch 1/2... Discriminator Loss: 1.0431... Generator Loss: 1.4336
Epoch 1/2... Discriminator Loss: 1.0715... Generator Loss: 2.1971
Epoch 1/2... Discriminator Loss: 0.8559... Generator Loss: 1.6649
Epoch 1/2... Discriminator Loss: 1.4629... Generator Loss: 0.5663
Epoch 1/2... Discriminator Loss: 1.0197... Generator Loss: 2.3181
Epoch 1/2... Discriminator Loss: 1.3370... Generator Loss: 0.6418
Epoch 1/2... Discriminator Loss: 0.8818... Generator Loss: 1.4240
Epoch 1/2... Discriminator Loss: 1.6460... Generator Loss: 0.4707
Epoch 1/2... Discriminator Loss: 1.9337... Generator Loss: 0.3501
Epoch 1/2... Discriminator Loss: 1.0291... Generator Loss: 1.1043

Epoch 1/2... Discriminator Loss: 1.3434... Generator Loss: 0.6728
Epoch 1/2... Discriminator Loss: 0.9033... Generator Loss: 1.3412
Epoch 1/2... Discriminator Loss: 1.5118... Generator Loss: 0.5001
Epoch 1/2... Discriminator Loss: 1.3226... Generator Loss: 0.7667
Epoch 1/2... Discriminator Loss: 0.9833... Generator Loss: 1.0875
Epoch 1/2... Discriminator Loss: 0.8792... Generator Loss: 1.4414
Epoch 1/2... Discriminator Loss: 0.9022... Generator Loss: 2.3660
Epoch 1/2... Discriminator Loss: 0.9310... Generator Loss: 1.8334
Epoch 1/2... Discriminator Loss: 0.9525... Generator Loss: 1.8847
Epoch 1/2... Discriminator Loss: 0.9145... Generator Loss: 2.0856

Epoch 1/2... Discriminator Loss: 1.5702... Generator Loss: 0.5224
Epoch 1/2... Discriminator Loss: 1.6709... Generator Loss: 0.4654
Epoch 1/2... Discriminator Loss: 1.0163... Generator Loss: 1.1649
Epoch 1/2... Discriminator Loss: 0.9700... Generator Loss: 1.0841
Epoch 1/2... Discriminator Loss: 1.0267... Generator Loss: 2.0636
Epoch 1/2... Discriminator Loss: 0.9675... Generator Loss: 1.4201
Epoch 1/2... Discriminator Loss: 1.1340... Generator Loss: 0.8231
Epoch 1/2... Discriminator Loss: 0.8418... Generator Loss: 1.6626
Epoch 1/2... Discriminator Loss: 1.3623... Generator Loss: 0.6419
Epoch 1/2... Discriminator Loss: 1.1561... Generator Loss: 0.8382

Epoch 1/2... Discriminator Loss: 0.9165... Generator Loss: 1.3181
Epoch 1/2... Discriminator Loss: 2.3313... Generator Loss: 0.3832
Epoch 1/2... Discriminator Loss: 0.9082... Generator Loss: 1.4724
Epoch 1/2... Discriminator Loss: 1.0026... Generator Loss: 1.0490
Epoch 1/2... Discriminator Loss: 1.7183... Generator Loss: 0.6498
Epoch 1/2... Discriminator Loss: 1.5474... Generator Loss: 0.5748
Epoch 1/2... Discriminator Loss: 0.9518... Generator Loss: 1.2638
Epoch 1/2... Discriminator Loss: 1.1974... Generator Loss: 0.8200
Epoch 1/2... Discriminator Loss: 1.1028... Generator Loss: 0.9384
Epoch 1/2... Discriminator Loss: 0.9618... Generator Loss: 1.2226

CelebA¶

Run your GANs on CelebA. It will take around 20 minutes on the average GPU to run one epoch. You can run the whole epoch or stop when it starts to generate realistic faces.

batch_size = 16
z_dim = 64
learning_rate = 0.0002
beta1 = 0.5


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 1

celeba_dataset = helper.Dataset('celeba', glob(os.path.join(data_dir, 'img_align_celeba/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, celeba_dataset.get_batches,
          celeba_dataset.shape, celeba_dataset.image_mode)

Epoch 1/1... Discriminator Loss: 2.6635... Generator Loss: 0.2675
Epoch 1/1... Discriminator Loss: 4.6404... Generator Loss: 0.0222
Epoch 1/1... Discriminator Loss: 0.5997... Generator Loss: 17.6954
Epoch 1/1... Discriminator Loss: 0.6490... Generator Loss: 5.3087
Epoch 1/1... Discriminator Loss: 0.5570... Generator Loss: 3.7793
Epoch 1/1... Discriminator Loss: 2.3933... Generator Loss: 0.1659
Epoch 1/1... Discriminator Loss: 0.5531... Generator Loss: 2.6736
Epoch 1/1... Discriminator Loss: 1.0832... Generator Loss: 1.4400
Epoch 1/1... Discriminator Loss: 0.9268... Generator Loss: 2.1777
Epoch 1/1... Discriminator Loss: 0.8227... Generator Loss: 1.3807

Epoch 1/1... Discriminator Loss: 1.1117... Generator Loss: 0.7185
Epoch 1/1... Discriminator Loss: 1.0948... Generator Loss: 0.8351
Epoch 1/1... Discriminator Loss: 0.6463... Generator Loss: 1.6834
Epoch 1/1... Discriminator Loss: 0.7976... Generator Loss: 1.2608
Epoch 1/1... Discriminator Loss: 0.5128... Generator Loss: 2.5563
Epoch 1/1... Discriminator Loss: 0.5777... Generator Loss: 2.5280
Epoch 1/1... Discriminator Loss: 0.8075... Generator Loss: 1.1300
Epoch 1/1... Discriminator Loss: 1.1061... Generator Loss: 1.5614
Epoch 1/1... Discriminator Loss: 0.8965... Generator Loss: 1.2147
Epoch 1/1... Discriminator Loss: 0.9944... Generator Loss: 2.7929

Epoch 1/1... Discriminator Loss: 0.7586... Generator Loss: 1.4039
Epoch 1/1... Discriminator Loss: 0.8851... Generator Loss: 1.0123
Epoch 1/1... Discriminator Loss: 0.6832... Generator Loss: 2.5324
Epoch 1/1... Discriminator Loss: 0.8814... Generator Loss: 1.3522
Epoch 1/1... Discriminator Loss: 1.1979... Generator Loss: 0.7788
Epoch 1/1... Discriminator Loss: 1.2079... Generator Loss: 1.2611
Epoch 1/1... Discriminator Loss: 0.7197... Generator Loss: 1.5255
Epoch 1/1... Discriminator Loss: 0.6785... Generator Loss: 1.5434
Epoch 1/1... Discriminator Loss: 0.9885... Generator Loss: 0.8975
Epoch 1/1... Discriminator Loss: 0.7476... Generator Loss: 1.3318

Epoch 1/1... Discriminator Loss: 0.6734... Generator Loss: 1.8879
Epoch 1/1... Discriminator Loss: 1.7991... Generator Loss: 3.9891
Epoch 1/1... Discriminator Loss: 1.6703... Generator Loss: 0.4212
Epoch 1/1... Discriminator Loss: 1.5421... Generator Loss: 0.5783
Epoch 1/1... Discriminator Loss: 1.5826... Generator Loss: 0.4993
Epoch 1/1... Discriminator Loss: 0.8584... Generator Loss: 1.4558
Epoch 1/1... Discriminator Loss: 1.1487... Generator Loss: 0.8784
Epoch 1/1... Discriminator Loss: 0.8865... Generator Loss: 1.9909
Epoch 1/1... Discriminator Loss: 0.8908... Generator Loss: 1.2086
Epoch 1/1... Discriminator Loss: 1.0031... Generator Loss: 0.9855

Epoch 1/1... Discriminator Loss: 1.0337... Generator Loss: 0.9865
Epoch 1/1... Discriminator Loss: 1.5870... Generator Loss: 0.4135
Epoch 1/1... Discriminator Loss: 1.2102... Generator Loss: 0.6687
Epoch 1/1... Discriminator Loss: 0.7906... Generator Loss: 1.7788
Epoch 1/1... Discriminator Loss: 2.1516... Generator Loss: 0.3743
Epoch 1/1... Discriminator Loss: 1.0663... Generator Loss: 0.6869
Epoch 1/1... Discriminator Loss: 1.1367... Generator Loss: 2.2093
Epoch 1/1... Discriminator Loss: 1.5369... Generator Loss: 0.4613
Epoch 1/1... Discriminator Loss: 0.8381... Generator Loss: 1.1308
Epoch 1/1... Discriminator Loss: 1.1380... Generator Loss: 0.8204

Epoch 1/1... Discriminator Loss: 1.1483... Generator Loss: 0.7581
Epoch 1/1... Discriminator Loss: 0.7018... Generator Loss: 1.6251
Epoch 1/1... Discriminator Loss: 0.6963... Generator Loss: 1.5230
Epoch 1/1... Discriminator Loss: 1.0761... Generator Loss: 0.8307
Epoch 1/1... Discriminator Loss: 0.9030... Generator Loss: 2.4963
Epoch 1/1... Discriminator Loss: 1.0924... Generator Loss: 0.8606
Epoch 1/1... Discriminator Loss: 1.4126... Generator Loss: 0.4693
Epoch 1/1... Discriminator Loss: 0.9107... Generator Loss: 1.2108
Epoch 1/1... Discriminator Loss: 1.0651... Generator Loss: 2.2131
Epoch 1/1... Discriminator Loss: 0.8908... Generator Loss: 0.9700

Epoch 1/1... Discriminator Loss: 1.5264... Generator Loss: 0.4727
Epoch 1/1... Discriminator Loss: 1.7077... Generator Loss: 0.3896
Epoch 1/1... Discriminator Loss: 0.8697... Generator Loss: 2.1902
Epoch 1/1... Discriminator Loss: 0.9070... Generator Loss: 1.2958
Epoch 1/1... Discriminator Loss: 0.9012... Generator Loss: 1.0022
Epoch 1/1... Discriminator Loss: 1.0713... Generator Loss: 0.7100
Epoch 1/1... Discriminator Loss: 1.1467... Generator Loss: 0.6772
Epoch 1/1... Discriminator Loss: 0.9658... Generator Loss: 1.3016
Epoch 1/1... Discriminator Loss: 1.0650... Generator Loss: 0.9418
Epoch 1/1... Discriminator Loss: 1.4278... Generator Loss: 0.4229

Epoch 1/1... Discriminator Loss: 0.8517... Generator Loss: 1.1015
Epoch 1/1... Discriminator Loss: 1.3034... Generator Loss: 0.5230
Epoch 1/1... Discriminator Loss: 1.4199... Generator Loss: 0.5978
Epoch 1/1... Discriminator Loss: 0.8662... Generator Loss: 1.5871
Epoch 1/1... Discriminator Loss: 1.2682... Generator Loss: 0.6300
Epoch 1/1... Discriminator Loss: 0.9830... Generator Loss: 1.1393
Epoch 1/1... Discriminator Loss: 0.7128... Generator Loss: 1.7024
Epoch 1/1... Discriminator Loss: 1.9617... Generator Loss: 0.2635
Epoch 1/1... Discriminator Loss: 1.2755... Generator Loss: 0.6425
Epoch 1/1... Discriminator Loss: 0.8810... Generator Loss: 1.7054

Epoch 1/1... Discriminator Loss: 0.6555... Generator Loss: 1.4767
Epoch 1/1... Discriminator Loss: 0.9380... Generator Loss: 0.9733
Epoch 1/1... Discriminator Loss: 0.9078... Generator Loss: 2.4473
Epoch 1/1... Discriminator Loss: 1.1310... Generator Loss: 1.2948
Epoch 1/1... Discriminator Loss: 1.0425... Generator Loss: 1.3108
Epoch 1/1... Discriminator Loss: 0.8350... Generator Loss: 1.5290
Epoch 1/1... Discriminator Loss: 0.9772... Generator Loss: 0.9077
Epoch 1/1... Discriminator Loss: 0.8249... Generator Loss: 1.5621
Epoch 1/1... Discriminator Loss: 0.6525... Generator Loss: 1.8978
Epoch 1/1... Discriminator Loss: 0.5849... Generator Loss: 1.6906

Epoch 1/1... Discriminator Loss: 0.8644... Generator Loss: 1.0419
Epoch 1/1... Discriminator Loss: 1.7353... Generator Loss: 0.3577
Epoch 1/1... Discriminator Loss: 0.9589... Generator Loss: 1.2316
Epoch 1/1... Discriminator Loss: 0.9459... Generator Loss: 0.9137
Epoch 1/1... Discriminator Loss: 0.9145... Generator Loss: 1.3813
Epoch 1/1... Discriminator Loss: 0.9237... Generator Loss: 1.1781
Epoch 1/1... Discriminator Loss: 0.9533... Generator Loss: 1.1184
Epoch 1/1... Discriminator Loss: 1.4142... Generator Loss: 0.4880
Epoch 1/1... Discriminator Loss: 0.8577... Generator Loss: 1.7742
Epoch 1/1... Discriminator Loss: 0.7904... Generator Loss: 1.3594

Submitting This Project¶

When submitting this project, make sure to run all the cells before saving the notebook. Save the notebook file as "dlnd_face_generation.ipynb" and save it as a HTML file under "File" -> "Download as". Include the "helper.py" and "problem_unittests.py" files in your submission.